This project explores the relationship between chess performance metrics and Elo ratings, aiming to determine if game performance can reliably predict player ratings or detect cheating.

Using a dataset of 1 million games from The Week in Chess, we leveraged Stockfish, a state-of-the-art chess engine, to analyze moves and compute metrics like Centipawn Loss and Winning Chance Loss.

Key questions include whether single-game metrics can effectively predict Elo and if aggregated metrics across multiple games serve as stronger predictors.

Our results showed weak correlation for single games but revealed stronger patterns when averaging metrics over multiple games. These findings offer valuable insights into Elo prediction and provide a foundation for developing improved anti-cheating tools and user-friendly applications.